Diagnostic development of damage of ship generator sets by means of electrical signals

The paper presents the use of selected signals to diagnose the development of damage to some components of generators driven by marine compression-ignition engines. The research was conducted under the natural operating conditions of vessels, usually during port stops. During the investigation, it was possible to observe apparent changes in the values of diagnostic symptoms, which corresponded to the development of gradual or sudden damage.The developing damages detected in the work were in metal components. Diagnostic signals for detecting developing defects measured the parameters of working and accompanying processes in various domains.The study sought proper signal analysis methods, so computer visualization and recording diagnostic parameters from various fields were used. Decimation, time domain windowing, wavelet statistics, color analysis, and machine learning were used for classification in artificial neural networks. Experimental studies demonstrated the ability to diagnose damage development processes. The advanced signal processing methods made it possible to obtain many signal measures, from which the most useful for diagnosis were selected.